private void writeSampleLikelihoods(
StringBuffer out, VariantContext vc, double[] log10Likelihoods) {
if (VQSRCalibrator != null) {
log10Likelihoods =
VQSRCalibrator.includeErrorRateInLikelihoods(VQSLOD_KEY, vc, log10Likelihoods);
}
double[] normalizedLikelihoods = MathUtils.normalizeFromLog10(log10Likelihoods);
// see if we need to randomly mask out genotype in this position.
for (double likeVal : normalizedLikelihoods) {
out.append(formatter.format(likeVal));
// out.append(String.format("%5.4f ",likeVal));
}
}
private Map<String, Object> calculateIC(final VariantContext vc) {
final GenotypesContext genotypes =
(founderIds == null || founderIds.isEmpty())
? vc.getGenotypes()
: vc.getGenotypes(founderIds);
if (genotypes == null || genotypes.size() < MIN_SAMPLES) return null;
int idxAA = 0, idxAB = 1, idxBB = 2;
if (!vc.isBiallelic()) {
// for non-bliallelic case, do test with most common alt allele.
// Get then corresponding indeces in GL vectors to retrieve GL of AA,AB and BB.
int[] idxVector = vc.getGLIndecesOfAlternateAllele(vc.getAltAlleleWithHighestAlleleCount());
idxAA = idxVector[0];
idxAB = idxVector[1];
idxBB = idxVector[2];
}
double refCount = 0.0;
double hetCount = 0.0;
double homCount = 0.0;
int N = 0; // number of samples that have likelihoods
for (final Genotype g : genotypes) {
if (g.isNoCall() || !g.hasLikelihoods()) continue;
if (g.getPloidy() != 2) // only work for diploid samples
continue;
N++;
final double[] normalizedLikelihoods =
MathUtils.normalizeFromLog10(g.getLikelihoods().getAsVector());
refCount += normalizedLikelihoods[idxAA];
hetCount += normalizedLikelihoods[idxAB];
homCount += normalizedLikelihoods[idxBB];
}
if (N < MIN_SAMPLES) {
return null;
}
final double p =
(2.0 * refCount + hetCount)
/ (2.0 * (refCount + hetCount + homCount)); // expected reference allele frequency
final double q = 1.0 - p; // expected alternative allele frequency
final double F = 1.0 - (hetCount / (2.0 * p * q * (double) N)); // inbreeding coefficient
Map<String, Object> map = new HashMap<String, Object>();
map.put(getKeyNames().get(0), String.format("%.4f", F));
return map;
}
/**
* Read in a list of ExactCall objects from reader, keeping only those with starts in startsToKeep
* or all sites (if this is empty)
*
* @param reader a just-opened reader sitting at the start of the file
* @param startsToKeep a list of start position of the calls to keep, or empty if all calls should
* be kept
* @param parser a genome loc parser to create genome locs
* @return a list of ExactCall objects in reader
* @throws IOException
*/
public static List<ExactCall> readExactLog(
final BufferedReader reader, final List<Integer> startsToKeep, GenomeLocParser parser)
throws IOException {
if (reader == null) throw new IllegalArgumentException("reader cannot be null");
if (startsToKeep == null) throw new IllegalArgumentException("startsToKeep cannot be null");
if (parser == null) throw new IllegalArgumentException("GenomeLocParser cannot be null");
List<ExactCall> calls = new LinkedList<ExactCall>();
// skip the header line
reader.readLine();
// skip the first "type" line
reader.readLine();
while (true) {
final VariantContextBuilder builder = new VariantContextBuilder();
final List<Allele> alleles = new ArrayList<Allele>();
final List<Genotype> genotypes = new ArrayList<Genotype>();
final double[] posteriors = new double[2];
final double[] priors = MathUtils.normalizeFromLog10(new double[] {0.5, 0.5}, true);
final List<Integer> mle = new ArrayList<Integer>();
final Map<Allele, Double> log10pNonRefByAllele = new HashMap<Allele, Double>();
long runtimeNano = -1;
GenomeLoc currentLoc = null;
while (true) {
final String line = reader.readLine();
if (line == null) return calls;
final String[] parts = line.split("\t");
final GenomeLoc lineLoc = parser.parseGenomeLoc(parts[0]);
final String variable = parts[1];
final String key = parts[2];
final String value = parts[3];
if (currentLoc == null) currentLoc = lineLoc;
if (variable.equals("type")) {
if (startsToKeep.isEmpty() || startsToKeep.contains(currentLoc.getStart())) {
builder.alleles(alleles);
final int stop = currentLoc.getStart() + alleles.get(0).length() - 1;
builder.chr(currentLoc.getContig()).start(currentLoc.getStart()).stop(stop);
builder.genotypes(genotypes);
final int[] mleInts = ArrayUtils.toPrimitive(mle.toArray(new Integer[] {}));
final AFCalcResult result =
new AFCalcResult(mleInts, 1, alleles, posteriors, priors, log10pNonRefByAllele);
calls.add(new ExactCall(builder.make(), runtimeNano, result));
}
break;
} else if (variable.equals("allele")) {
final boolean isRef = key.equals("0");
alleles.add(Allele.create(value, isRef));
} else if (variable.equals("PL")) {
final GenotypeBuilder gb = new GenotypeBuilder(key);
gb.PL(GenotypeLikelihoods.fromPLField(value).getAsPLs());
genotypes.add(gb.make());
} else if (variable.equals("log10PosteriorOfAFEq0")) {
posteriors[0] = Double.valueOf(value);
} else if (variable.equals("log10PosteriorOfAFGt0")) {
posteriors[1] = Double.valueOf(value);
} else if (variable.equals("MLE")) {
mle.add(Integer.valueOf(value));
} else if (variable.equals("pNonRefByAllele")) {
final Allele a = Allele.create(key);
log10pNonRefByAllele.put(a, Double.valueOf(value));
} else if (variable.equals("runtime.nano")) {
runtimeNano = Long.valueOf(value);
} else {
// nothing to do
}
}
}
}
